Kevin Enderby [Mon, 26 Sep 2016 21:11:03 +0000 (21:11 +0000)]
Next set of additional error checks for invalid Mach-O files for the
other load commands that use the Mach::linkedit_data_command type
but not used in llvm libObject code but used in llvm tool code.
This includes LC_FUNCTION_STARTS, LC_SEGMENT_SPLIT_INFO
and LC_DYLIB_CODE_SIGN_DRS load commands.
Piotr Padlewski [Mon, 26 Sep 2016 20:37:32 +0000 (20:37 +0000)]
[thinlto] Basic thinlto fdo heuristic
Summary:
This patch improves thinlto importer
by importing 3x larger functions that are called from hot block.
I compared performance with the trunk on spec, and there
were about 2% on povray and 3.33% on milc. These results seems
to be consistant and match the results Teresa got with her simple
heuristic. Some benchmarks got slower but I think they are just
noisy (mcf, xalancbmki, omnetpp)- running the benchmarks again with
more iterations to confirm. Geomean of all benchmarks including the noisy ones
were about +0.02%.
I see much better improvement on google branch with Easwaran patch
for pgo callsite inlining (the inliner actually inline those big functions)
Over all I see +0.5% improvement, and I get +8.65% on povray.
So I guess we will see much bigger change when Easwaran patch will land
(it depends on new pass manager), but it is still worth putting this to trunk
before it.
Implementation details changes:
- Removed CallsiteCount.
- ProfileCount got replaced by Hotness
- hot-import-multiplier is set to 3.0 for now,
didn't have time to tune it up, but I see that we get most of the interesting
functions with 3, so there is no much performance difference with higher, and
binary size doesn't grow as much as with 10.0.
Statistic: Only print statistics on exit for -stats
Previously enabling the statistics with EnableStatistics() would lead to
them getting printed to stderr/-info-output-file on exit. However
frontends may want a way to enable statistics and do the printing on
their own instead of the forced printing on exit.
This changes the code so that only the -stats option enables printing on
exit, EnableStatistics() only enables the tracking but requires invoking
one of the PrintStatistics() variants.
[llvm-cov] Silence a warning from the MSVC runtime (NFC)
Rework getLongestCommonPrefixLen() so that it doesn't access string null
terminators. The old version with std::mismatch would do this:
|
v
Strings[0] = ['a', nil]
Strings[1] = ['a', 'a', nil]
^
|
This should silence a warning from the MSVC runtime (PR30515). As
before, I tested this out by preparing a coverage report for FileCheck.
Thanks to Yaron Keren for the report!
Matthew Simpson [Mon, 26 Sep 2016 17:08:37 +0000 (17:08 +0000)]
[LV] Scalarize instructions marked scalar after vectorization
This patch ensures that we actually scalarize instructions marked scalar after
vectorization. Previously, such instructions may have been vectorized instead.
Gor Nishanov [Mon, 26 Sep 2016 15:49:28 +0000 (15:49 +0000)]
[Coroutines] Part14: Handle coroutines with no suspend points.
Summary:
If coroutine has no suspend points, remove heap allocation and turn a coroutine into a normal function.
Also, if a pattern is detected that coroutine resumes or destroys itself prior to coro.suspend call, turn the suspend point into a simple jump to resume or cleanup label. This pattern occurs when coroutines are used to propagate errors in functions that return expected<T>.
Don't match the UXTW extended reg forms of ADD/ADDS/SUB/SUBS if the
32-bit to 64-bit zero-extend can be done for free by taking advantage
of the 32-bit defining instruction zeroing the upper 32-bits of the X
register destination. This enables better instruction selection in a
few cases, such as:
sub x0, xzr, x8
instead of:
mov x8, xzr
sub x0, x8, w9, uxtw
Add support to optionally limit the size of jump tables.
Many high-performance processors have a dedicated branch predictor for
indirect branches, commonly used with jump tables. As sophisticated as such
branch predictors are, they tend to have well defined limits beyond which
their effectiveness is hampered or even nullified. One such limit is the
number of possible destinations for a given indirect branches that such
branch predictors can handle.
This patch considers a limit that a target may set to the number of
destination addresses in a jump table.
Patch by: Evandro Menezes <e.menezes@samsung.com>, Aditya Kumar
<aditya.k7@samsung.com>, Sebastian Pop <s.pop@samsung.com>.
The index of the new insertelement instruction was evaluated in the
wrong way, it was considered as the index of the inserted value instead
of index of the position, where the value should be inserted.
Andrea Di Biagio [Mon, 26 Sep 2016 12:07:23 +0000 (12:07 +0000)]
[InstCombine] Teach the udiv folding logic how to handle constant expressions.
This patch fixes PR30366.
Function foldUDivShl() worked under the assumption that one of the values
in input to the function was always an instance of llvm::Instruction.
However, function visitUDivOperand() (the only user of foldUDivShl) was
clearly violating that precondition; internally, visitUDivOperand() uses pattern
matches to check the operands of a udiv. Pattern matchers for binary operators
know how to handle both Instruction and ConstantExpr values.
This patch fixes the problem in foldUDivShl(). Now we use pattern matchers
instead of explicit casts to Instruction. The reduced test case from PR30366
has been added to test file InstCombine/udiv-simplify.ll.
James Molloy [Mon, 26 Sep 2016 07:26:24 +0000 (07:26 +0000)]
[ARM] Promote small global constants to constant pools
If a constant is unamed_addr and is only used within one function, we can save
on the code size and runtime cost of an indirection by changing the global's storage
to inside the constant pool. For example, instead of:
This can cause significant code size savings when many small strings are used in one
function (4 bytes per string).
This recommit contains fixes for a nasty bug related to fast-isel fallback - because
fast-isel doesn't know about this optimization, if it runs and emits references to
a string that we inline (because fast-isel fell back to SDAG) we will end up
with an inlined string and also an out-of-line string, and we won't emit the
out-of-line string, causing backend failures.
It also contains fixes for emitting .text relocations which made the sanitizer
bots unhappy.
[X86] Optimization for replacing LEA with MOV at frame index elimination time
Summary:
Replace a LEA instruction of the form 'lea (%esp), %ebx' --> 'mov %esp, %ebx'
MOV is preferable over LEA because usually there are more issue-slots available to execute MOVs than LEAs. Latest processors also support zero-latency MOVs.
[PM] Refactor this unittest a bit to remove duplicated code. This was
suggested at one point during code review and I deferred it to
a follow-up commit.
[PM] Add a unittest for invalidating module analyses with an SCC pass.
This reinstates r280447. Original commit log:
This wasn't really well explicitly tested with a nice unittest before.
It seems good to have reasonably broken out unittests for this kind of
functionality as I'm workin go other invalidation features to make sure
none of the existing ones regress.
This still has too much duplicated code, I plan to factor that out in
a subsequent commit to use common helpers for repeated parts of this.
Sanjoy Das [Mon, 26 Sep 2016 02:44:10 +0000 (02:44 +0000)]
[SCEV] Assign LoopPropertiesCache in the move constructor
In a previous change I collapsed two different caches into one. When
doing that I noticed that ScalarEvolution's move constructor was not
moving those caches.
To keep the previous change simple, I've moved that bugfix into this
separate change.
Sanjoy Das [Mon, 26 Sep 2016 02:44:07 +0000 (02:44 +0000)]
[SCEV] Combine two predicates into one; NFC
Both `loopHasNoSideEffects` and `loopHasNoAbnormalExits` involve walking
the loop and maintaining similar sorts of caches. This commit changes
SCEV to compute both the predicates via a single walk, and maintain a
single cache instead of two.
[AVX-512] Remove duplicate instructions for converting integer to scalar floating point. We can use patterns to point to the other instructions instead.
Add some predicated searching functions to StringRef.
This adds 4 new functions to StringRef, which can be used to
take or drop characters while a certain condition is met, or
until a certain condition is met. They are:
take_while - Return characters until a condition is not met.
take_until - Return characters until a condition is met.
drop_while - Remove characters until a condition is not met.
drop_until - Remove characters until a condition is met.
Internally, all of these functions delegate to two additional
helper functions which can be used to search for the position
of a character meeting or not meeting a condition, which are:
find_if - Find the first character matching a predicate.
find_if_not - Find the first character not matching a predicate.
[AVX-512] Don't use two opcodes for INTR_TYPE_SCALAR_MASK_RM. The handling was such that if the second opcode was present the first was ingored, so we can just have one opcode.
[X86] Teach combineShuffle to avoid creating floating point operations with integer types and integer operations with floating point types. Seems isOperationLegal lies for mismatched types and operations.
Stop looking at users of UndefValue and ConstantPointerNull in the
objective C ARC optimizers. The other users aren't actually
interesting, since they're not pointing at a particular object. I
imagine these calls could be optimized through -instcombine... maybe
they already are?
These early returns will be required at some point in the future, with a
WIP patch that asserts when someone accesses a use-list on ConstantData.
Analysis: Return early for UndefValue in computeKnownBits
There is no benefit in looking through assumptions on UndefValue to
guess known bits. Return early to avoid walking their use-lists, and
assert that all instances of ConstantData are handled here for similar
reasons (UndefValue was the only integer/pointer holdout).
Assumptions on UndefValue and ConstantPointerNull aren't relevant to
other users. Ignore them entirely to avoid wasting cycles walking
through their (possibly extremely extensive (cross-module)) use-lists.
It wasn't clear how to add a specific test for this, and it'll be
covered anyway by an eventual patch that asserts when trying to access
the use-list of an instance of ConstantData.
Analysis: Return early in isKnownNonNullAt for ConstantData
Check and return early for ConstantPointerNull and UndefValue
specifically in isKnownNonNullAt, and assert that ConstantData never
make it to isKnownNonNullFromDominatingCondition.
This confirms that isKnownNonNullFromDominatingCondition never walks
through the use-list of an instance of ConstantData. Given that such
use-lists cross module boundaries, it never really made sense to do so,
and was potentially very expensive.
[RegisterBankInfo] Document the design choice for the BreakDown.
This adds a comment explaining why we will duplicate PartialMapping to
represent the breakdown for complex mappings (mappings with more than
one partial mapping), instead of using an array of pointer.
This is a step toward statically allocate ValueMapping. Like the
previous few commits, the goal is to move toward a TableGen'ed like
structure with no dynamic allocation at all.
[RegisterBankInfo] Keep valid pointers for PartialMappings.
Previously we were using the address of the unique instance of a partial
mapping in the related map to access this instance. However, when the
map grows, the whole set of instances may be moved elsewhere and the
previous addresses are not valid anymore.
Instead, keep the address of the unique heap allocated instance of a
partial mapping.
Note: I did not see any actual bugs for that problem as the number of
partial mappings dynamically allocated is small (<= 4).
GlobalStatus: Don't walk use-lists of ConstantData
Return early from llvm::isSafeToDestroyConstant() whenever the value
`isa<ConstantData>()`. These constants are shared across the
LLVMContext. We never really want to delete them here, and walking
their use-lists can be very expensive.
(This is motivated by an eventual goal of removing use-lists entirely
from ConstantData.)
[x86] fix FCOPYSIGN lowering to create constants instead of ConstantPool loads
This is similar to:
https://reviews.llvm.org/rL279958
By not prematurely lowering to loads, we should be able to more easily eliminate
the 'or' with zero instructions seen in copysign-constant-magnitude.ll.
We should also be able to extend this code to handle vectors.
The NativeObjectOutput class has a design problem: it mixes up the caching
policy with the interface for output streams, which makes the client-side
code hard to follow and would for example make it harder to replace the
cache implementation in an arbitrary client.
This change separates the two aspects by moving the caching policy
to a separate field in Config, replacing NativeObjectOutput with a
NativeObjectStream class which only deals with streams and does not need to
be overridden by most clients and introducing an AddFile callback for adding
files (e.g. from the cache) to the link.
Teresa Johnson [Fri, 23 Sep 2016 20:35:19 +0000 (20:35 +0000)]
[gold] Split plugin options controlling ThinLTO and codegen parallelism.
Summary:
As suggested in D24826, use different options for ThinLTO backend
parallelism from the option controlling regular LTO code gen
parallelism. They are already split in the LTO API, and this enables
controlling them with different clang options.
[llvm-cov] Get rid of all invalid filename references
We used to append filenames into a vector of std::string, and then
append a reference to each string into a separate vector. This made it
easier to work with the getUniqueSourceFiles API. But it's buggy.
std::string has a small-string optimization, so you can't expect to
capture a reference to one if you're copying it into a growing vector.
Add a test that triggers this invalid reference to std::string scenario,
and kill the issue with fire by just using ArrayRef<std::string>
everywhere.
projects / llvm / log